Cost-Utility of Three Approaches to Sleep Apnea: Polysomnography, Home Testing, Empirical Therapy
Abstract & Commentary
Synopsis: The advantage gained by polysomnography over other approaches was comparable to that of other procedures like coronary artery bypass surgery, dialysis, and screening asymptomatic patients for carotid artery stenosis.
Source: Chervin RD, et al. Ann Intern Med 1999;130:496-505.
Obstructive sleep apnea syndrome (osas) occurs with repeated obstruction of breathing during sleep and is associated with daytime sleepiness, significant cardiovascular morbidity, and increased mortality. Recent data suggest that 2-4% of adults have this disorder. Nasal continuous positive airway pressure (NCPAP) is the most commonly used treatment and has a favorable cost-utility ratio. However, OSAS remains undiagnosed in at least 82% of men and 93% of women with the condition. The gold standard for the diagnosis of OSAS is nocturnal polysomnography, but this is associated with significant expense, at a cost of $1000-1400 per patient. Some investigators have suggested that sleep apnea can be diagnosed with either a home study or by clinical symptoms and physical finding alone at a much lower cost, albeit with less accuracy. The present study was conducted to compare the cost-effectiveness of these strategies using decision analysis techniques.
Chervin and associates first constructed a decision tree in which a hypothetical adult with OSAS was evaluated and treated. Within the decision tree, three alternative approaches to diagnosis were used, chances of positive or negative test results were considered, the possibilities of treatment with or without NCPAP were entered, the probability that OSAS was actually present was entered, and the resulting health state was considered. The data used in the tree were derived from consecutive, sleep center-referred, adult case series with the diagnosis of OSAS. On average, the patients modeled were in the sixth decade of life, most were males, and many had cardiovascular comorbidity.
Outcome measures included the quality-adjusted life years (QALY) (utility × expected life span) for the first five years (QALY 5) and incremental cost. A decision analysis was used to determine the magnitude of the advantage afforded by the best diagnostic approach—polysomnography, home study, or no testing—measured in QALY 5. Also, five-year incremental cost-utility ratios in U.S. dollars of obtaining that advantage were calculated. After derivation of results of a base case, a univariate sensitivity analysis was done to test the effects on the model of different but plausible utilities, survival rates, pretest probabilities of OSAS, test characteristics, and costs. Baseline utilities of four possible health states (outcomes), mean survival during five-year period after diagnostic evaluation and resulting QALY were calculated. QALY were: 4.104 for patients with OSAS using NCPAP; 2.972 for patients using NCPAP without OSAS; 2.804 for patients with OSAS not using NCPAP; and 3.538 for patients without OSAS and not using NCPAP. Mean survival during the five-year period was 5, 5, 4.7, and 5 years, respectively.
An Eden trace model 2700 with a sensitivity and specificity of 0.95 and 0.96 (respectively) in polysomnographically proven OSAS was used in the model. The proportion of all home studies expected to yield positive results was calculated to be 0.81. In no test branch did all patients with OSAS receive treatment.
Charges as a proxy of cost were used to calculate the incremental cost. The average total charge for patients in any given branch of the decision tree was calculated, including five-year total charges for any tests, office visits, and NCPAP setups, along with the probability that each charge would be incurred. Charges after the first year were discounted at the rate of 3% per year including the future utilities. In a baseline decision analysis, polysomnography generated higher QALY 5 (4.019) and diagnostic and treatment charges ($4210) than home study (3.955 and $3460, respectively) and no testing (3.934 and $3020, respectively). The incremental cost-utility ratio for polysomnography compared with home study and no testing were $13,431 and $9165, respectively, per QALY gained.
A multivariate analysis was then done to recalculate the cost-utility ratios under the assumption that the frequency of OSAS was only 0.35 and that the charge for a home study was only $50. Under these assumptions, the cost-utility ratio for polysomnography compared with home study increased to $30,070 per additional QALY 5 gained by polysomnography if the sensitivity and specificity of the test remained at 95% and 96%, respectively. The cost-utility ratio would decrease to $40,000 if the sensitivity and specificity dropped to 80% and 70%, respectively.
Comment by David Ost, MD
Because of increasing awareness of OSAS among physicians and the lay public, the importance of accurate diagnosis and treatment is increasing. Almost every insurance company reimburses for standard overnight polysomnography for the diagnosis of OSAS. Because of its expense (one night polysomnography costs approximately $1000-4000), other inexpensive diagnostic modalities are being used in clinical practice, including home studies. These studies are cheap and easy to use, but their overall effect on making an accurate diagnosis and long-term costs is still questionable. The present study was conducted to determine the cost-utility of three approaches to OSAS: polysomnography, home testing, and empirical therapy.
Chervin et al demonstrate that formal polysomnography results in an improved quality of life at a cost that is comparable to and less than many other well accepted therapeutic interventions. Importantly, this result was true over a wide range of potential values for sensitivity, specificity, and prevalence of disease. This study provides important information to the clinician for evaluating the many patients with possible OSAS. Cheaper but less accurate home tests may not be better. Clearly, this study may also have a tremendous effect in terms of public health policy and financing.
Which of the following were used in the decision tree constructed by Chervin et al?
a. Chances of positive or negative test results
b. The possibilities of treatment with or without NCPAP
c. The probability that OSAS was actually present
d. The resulting health state
e. All of the above